Women’s right to property and the child quantity-quality trade-off: evidence from India

We study the effects of a series of state and federal reforms that granted Indian women equal inheritance rights on the quantity and quality of children. Using a difference-in-differences methodology, we find that women who were affected by the state reforms had 0.4 more children. State reforms did not have any effect on children’s heights. To assess the impact of the federal reform, we use panel data on women and a novel treatment based on the timing of their fathers’ deaths. We find that women affected by the reform had on average 0.22 fewer children and had taller children on average. While the federal reform had no effect on the number of daughters born to this group, the number of sons born declined. Thus, we see evidence that granting property rights to women could potentially impact fertility decisions, children’s health outcomes, and gender imbalance.


Introduction
The issue of quantity and quality of children is salient for developing countries in Asia and Africa, which have historically seen high population growth with low levels of human development.These countries see severe stunting in children and skewed sex ratios.Policies for empowering women, such as granting property rights through land redistribution and inheritance, can have profound implications for societies through changes in fertility outcomes, nutritional investments in children, and gender imbalance.
In this paper, we compare and test whether granting land and property inheritance rights to women in India leads to a quantity-quality trade-off under two different inheritance reform regimes.We also test whether these reforms have differential effects on male and female children.
We examine the impact of reforms in four states between 1986-94 and a landmark federal law, the Hindu Succession Amendment Act of 2005, that granted Hindu women equal inheritance rights to their natal joint family properties.1 Specifically, we investigate whether the amendments resulted in treated women having fewer children and the children born to these women having higher height-for-age z-scores-a measure of quality.We make use of rich panel data from the India Human Development Survey (IHDS), which records information on 42,152 households in two waves: 2004-05 and 2011-12.
To examine the impact of the state reforms, we use a difference-in-differences methodology exploiting variation in the state of residence and marital status of Hindu women.This closely follows the empirical strategy used by Bose and Das (2020).We see whether the state amendment had an effect on the number of children born to treated women and on the height-for-age z-score of children born to these women.We find that women who were eligible for an equal share in their ancestral property under the state amendments had on average 0.4 more children.Households belonging to income quartiles below the median had the strongest effect.There was no effect on the quality of the children born.The average number of sons born to women in this group increased by 0.26, while the number of daughters born increased by 0.2.
To analyze the effect of the 2005 federal amendment, we construct a panel of women across the two IHDS waves.We use the treatment of the death of the father between the two waves to observe changes in the number of children born and changes in the height-for-age z-score of the child.We find that women whose fathers died after 2005 had on average 0.15 fewer children and children born to such women were taller by 2.28 standard deviation.We thus see evidence of a quantity-quality trade-off.While the federal amendment had no effect on the number of daughters born, the number of sons born declined by 0.14.These results provide suggestive evidence of weakening son preference.Our results for the state-level and federal amendments continue to hold when using probability of birth as a measure of fertility and weight-for-age z-score as a measure of quality.
We propose that the amendments impacted outcomes for women and children through three different channels.The first is a direct positive wealth shock from inheriting property.However, given the divergence between de jure and de facto implementation of the reforms, the likelihood of actually inheriting property may be low.Hence, a positive wealth shock can only partially explain our results.The second channel is through compensating mechanisms such as education or other resources that women may receive in lieu of property which could empower them to realize their desired fertility outcomes.The third channel is an increase in women's intrahousehold bargaining power because of the expectation that women could potentially inherit property or other resources from their natal household in the future due to the legislation.
Aside from the differences in identification strategies, sample sizes, and the nature of the state and federal amendments (the former was applicable only to unmarried women at the time of the reform while the latter applied to all women), the divergence in the effects of the state and federal amendments can be attributed to the following reasons.First, existing theoretical frameworks imply that the effects on quantity could go in either direction (Rosenzweig and Evenson 1977;Becker and Lewis 1973).Second, access to prenatal sex determination became easier around the time of the federal amendment.Thus, while women were able to achieve their desired fertility outcomes due to the state amendments by using a stopping rule of having more children until the desired number of sons was achieved, women potentially affected by the federal amendment possibly used prenatal sex determination and abortion procedures to attain their desired fertility outcomes.On the other hand, the fact that we see the number of sons decline due to the federal amendment possibly points to changing social norms and expectations from children, especially sons, for old age security in India.
This paper contributes to three strands of literature.The first strand examines the impact of women's economic empowerment on fertility decisions and outcomes for children in the context of developing countries.Theory suggests that the impact would differ based on whether the channel is through an income effect or a wealth effect.If we take into account women's participation in the workforce, then given the opportunity cost of the time taken to raise children, an increase in women's incomes may be accompanied by a fall in the quantity of children but a rise in quality since the additional income is invested in the children (Becker 1965;Black et al. 2013;Hotz and Miller 1988;Willis 1973).Schultz (1985) and Heckman and Walker (1990) provide evidence of this for developed countries.Rosenzweig and Evenson (1977) find that an increase in women's wages reduced family size and increased school enrollment rates in India.Jensen (2012) also finds that women in rural India who entered the workforce reported a desire to have fewer children.In the neoclassical model first proposed by Becker (1960), children are considered to be a normal good-families that see a rise in incomes decide to have more children.There is some empirical evidence of this in the USA, especially in the case of a rise in men's income or household wealth (Black et al. 2013;Dettling and Kearney 2014;Lovenheim and Mumford 2013).Thus, a wealth effect could result in an increase in the number of children.However, the effect of an increase in women's wealth on fertility is somewhat ambiguous.If this increase improves bargaining power for women within the household, they would be better able to attain their desired fertility levels.Studies show that granting property and land inheritance rights to women has yielded benefits such as improvement in health outcomes for young children (Allendorf 2007) and greater participation of women in household decision making, including fertility decisions (Ali et al. 2015;Harari 2019;Mookerjee 2019).If women's preference is for fewer children, we could see a decline in fertility rates with improved health outcomes as more resources are allocated to them.Eswaran (2002) posits that if the primary motivation to have children is old age security and if the cost of raising children is borne disproportionately by women, an increase in women's autonomy and bargaining power within the household would result in reduced fertility and increased investment in children's healthcare.Doepke and Tertilt (2018) show that women's empowerment is associated with lower fertility in the context of developing countries.
A set of studies examine the role of son preference in shaping fertility decisions as well as the effect of policies and economic empowerment on son preference.Studies show that women use a stopping rule-the practice of having children until the desired number of sons is reached (Bhalotra and Van Soest 2008;Bose and Das 2020;Clark 2000;Rosenblum 2013).Studies also show that son preference has resulted in skewed sex ratios (Anukriti et al. 2022;Bhalotra and Cochrane 2010).Distinguishing between absolute and relative wealth, Gaudin (2011) finds that an increase in the former is associated with lower son preference in India.In certain cases, policies that empower women may weaken son preference and improve survival rates of girl children (see Qian 2008).Our paper contributes to this body of work by examining the relationship between property rights for women and son preference.
The Hindu Succession Act of 1956 in India excluded women from inheriting a share in the ancestral property.A series of state amendments granted equal inheritance rights to Hindu women.2A number of studies have examined the impact of these reforms on outcomes for women and children.Deininger et al. (2013) show that the reforms led to a rise in levels of primary education attained by girls.Roy (2015) finds that while girls in the treatment group are not more likely to inherit land relative to girls in the control group, they had better educational outcomes.Thus, parents may seek to compensate unmarried girls for gifting ancestral property to sons by investing in their education.For girls of marriageable age, there was a rise in the amount of dowry payments.Deininger et al. (2019) find that while the number of daughters born to women who inherited property declined, the daughters had higher survival rates.Bhalotra et al. (2020) and Rosenblum (2015) show that state-level amendments had the unintended consequence of increasing female mortality.A study by Agarwal et al. (2021) examines land ownership among a panel of 183 women across nine Indian states and finds that widows are more likely to own land.This implies that land is more likely to be passed on to women from the marital family rather than the natal family.This is the third strand of literature that we contribute to.To the best of our knowledge, no published study has examined the effect of the federal amendment to the Hindu Succession Act in India using the father's death as an identification strategy.This is also the first paper to study and compare the effects of both state and federal amendments.
The rest of the paper proceeds as follows.Section 2 explains the nature of inheritance laws and amendments in India.Section 3 discusses the data.Section 4 describes the empirical strategy.Section 5 presents the results and robustness checks, discusses the mechanisms driving the results, and provides potential reasons for divergence in the results for the state and federal amendments.Section 6 concludes.

Background
The Hindu Succession Act (HSA) of 1956 codified Hindu inheritance practices into law in India. 3The law governs Hindu intestate succession-that is, succession in the absence of a will.Under this Act, property was classified as "joint," "separate," and "self acquired."Joint family property is acquired by inheritance from the paternal ancestor (Agarwal 1994).The Act also defined coparceners as those having a right to ancestral or joint property by birth and to demand partition of said property.In the HSA 1956, this right was restricted to male members of the joint family; women were not given coparcenary rights to the joint family property.The law applied to all Indian states except Jammu and Kashmir. 4etween 1976 and 1994, five states passed amendments to reform the HSA-Kerala in 1976, Andhra Pradesh in 1986, Tamil Nadu in 1989, and Maharashtra and Karnataka in 1994.In Kerala, the institution of joint family property was abolished.In the remaining states of Karnataka, Andhra Pradesh, Maharashtra, and Tamil Nadu, daughters were given right to their father's ancestral property upon his death intestate.However, only women who were unmarried at the time of the reform were eligible to inherit property.
The Hindu Succession Amendment Act in 2005 is a landmark pan-India inheritance reform that grants coparcenary rights to Hindu women.It was introduced by the 174th Report of Law Commission in 2000 and came into effect in September 2005. 5By allowing all women to claim equal share to their joint family property, it equalized inheritance practices across gender in India.It was applicable only to women whose fathers were alive as of 9 September 2005. 6he 2005 amendment was applicable to both married and unmarried women.This is a key difference between the state amendments and the federal amendment.As per the Indian Constitution, the federal amendment overrides earlier state amendments. 7ence, all women in the five early reform states became eligible to be coparceners as of 2005 regardless of marital status.
We perused legal cases and judgements to see whether the state and federal amendments were indeed effective in granting coparcenary rights to women.The following case is worth highlighting.In S Sai Reddy v. S Narayan Reddy & Ors (1991), two unmarried daughters approached the High Court of Andhra Pradesh to claim an equal share in their ancestral property.The family members had an understanding that the property was to be divided among the male members only.This provisional agreement took place before the state Amendment Act was passed in Andhra Pradesh.After the Amendment was passed, the daughters sought to also get an equal share in the property.The High Court ruled in favor of the daughters.The Court also noted that since the Act was implemented with the avowed objective of rectifying the unequal treatment of women, it should be interpreted in a liberal spirit.
Subsequent judgements by courts have had the effect of further strengthening the inheritance rights of women.In Mrs Sujata Sharma v. Shri Manu Gupta (2015), the High Court of the state of Delhi held that under the federal Amendment Act of 2005, the eldest woman member had not just coparcenary rights but also the right to be the manager of the ancestral property.

Data
We use the India Human Development Survey (IHDS), conducted by the University of Maryland and the National Council of Applied Economic Research.The IHDS is a panel dataset recording information on 42,152 households across villages and towns in India.The IHDS was conducted in two waves: 2004-05 and 2011-12, interviewing the same households at a rate of 85% across waves.The IHDS collects information on all states and union territories except the Andaman and Nicobar Islands and Lakshadweep.In our study, we limit our analysis to 16 large states (See Fig. 1). 8he IHDS provides a comprehensive database to conduct our analysis for the following reasons.It has information about land ownership by households, which allows us to construct a robust control vector for the state reform analysis and conduct analysis with subsamples of landowning households.Second, the timing of the first wave of the survey optimizes how much data is available to study the impact of state-level reforms in the 1980 s and 1990 s without allowing the 2005 federal reform to bias our results.Finally, the panel allows us to track fertility changes and death of fathers between two survey waves.We are able to exploit the information on the death of the fathers of Fig. 1 States considered for analysis.Note: Between 1976 and 1994, five Indian states passed amendments granting Hindu women who were unmarried at the time of the amendment the right to inherit an equal share of the ancestral property if the father died intestate.While five states passed amendments, our paper examines the effect in four of these states viz.Maharashtra, Karnataka, Tamil Nadu, and Andhra Pradesh.We exclude Kerala from this study because it abolished joint family property ownership.The amendment of the federal law in 2005 applied to all states.We restrict our analysis to 16 states, including the early four reform states.Source: Authors' own creation women to estimate the effect of the federal law on fertility outcomes of women and height-for-age z-score of children born to treated women.
The IHDS surveys individuals within households on health and socio-economic indicators.Data on family structure is collected by recording the relationships of individuals to each other and to the head of the household.The individual survey also records data on the status (whether living in same household, living in a different household, or dead) of the father of the respondent.This allows us to know if the father died across survey waves.
The survey also records information on individuals termed as "eligible women"; these are married women of ages 15-49 in the first wave of data collection.This information includes the women's socioeconomic indicators, health outcomes of the woman and her children, and other relevant data about intra-household gender relations.Since our paper studies women's fertility behaviors, information about use of contraceptives and abortions and number of women in a household allow us to test the impact of female decision making power on fertility.The data consists of 50,958 observations, with 25,479 eligible women interviewed in each wave.
The outcome variables are recorded as part of the eligible women's questionnaire.The first outcome variable of interest is the number of children born to a woman.The number of sons and daughters are also recorded, which allows us to analyze fertility by gender.The outcome variable for quality is the height-for-age z-score of the child.

Women's data
For the analysis on the impact of state and federal amendments on the number of children, the unit of observation is the household's eligible woman.We use data from the first wave (IHDS 1) to test the impact of state amendments to the HSA 1956 legislated between 1986 and 1994.We analyze the impact of state-level reforms using a sample 11,624 women surveyed in IHDS 1 for whom we had details on a number of variables including contraception and abortion.The mean number of children born to these women is 2.6 (see panel A in Table 1).
Analysis of the federal reform involved tracking the same woman across two waves and observing changes in the fertility rates of treatment and control groups.The IHDS is a household level panel, and creating an individual panel from this involves carefully identifying the same individuals within households across surveys.For some households, the IHDS does not track the same eligible women across two waves.To construct a panel, we tracked women that had the same inter-household and intrahousehold identifiers and year-of-birth across two waves.This allows us to identify a sample of 5301 eligible women. 9he IHDS records whether the father of the respondent is living in the same household, whether he is living in a different household, and whether he has died in a separate field (which the questionnaire notes as "father's status").This allows us to track women whose fathers died between two waves of survey.Our final sample consists of Hindu women for whom information on their education, general health, abortion, contraceptive use, father's status, spouse's education, share of women in the household, and household income was available. 10To verify whether the father's status was recorded without errors, we compared the response recorded in the second wave of the survey (IHDS 2) to the response recorded to the question: "Is your father alive" that was asked to the women in the second wave.11For 37 women, the responses recorded in the father's status were different from the response recorded under "Is your father alive."12These were dropped from our sample.We also drop women who were above the age of 50 in the second wave of IHDS from our sample. 13The final sample con- sists of 378 Hindu women that were interviewed across the two waves.Among these women, 202 reported that their father died between the two waves, and they represent our treatment group.Summary statistics for the pooled sample of women are provided in Table 2.

Children's data
For the analysis on the impact of the reforms on the quality of children, the unit of observation is the child born to the household's eligible woman.
To analyze the quality of children born to women impacted by state reforms, we matched children in households to eligible women interviewed in the first wave.We identified children by restricting our data to household members aged four and under since studies show that early childhood nutrition affects outcomes like stunting for children (see Glover-Amengor et al. 2016;Tarozzi and Mahajan 2007).We limited Including women above 50 in the sample would create endogeneity concerns since fathers of older women are more likely to die and the older women would have already attained their fertility outcomes before the first wave of IHDS.123 our sample to households that had only one married woman.This is to reasonably ensure that all the children in the household were born to the same woman. 14 We identify a treatment group of children born to women who lived in states that amended the law and were unmarried at the time of legislation of the state reform.Our main outcome variable is the height-for-age z-score of the child. 15After dropping observations that had no information on the outcome variable or on the control vector of woman and child characteristics, we get a sample of 1632 children.
For the analysis of the federal amendment's impact on the quality of children, we matched children across the two waves using inter-household and intra-household identifiers.We then combined the children's panel with the eligible women's panel. 16 We dropped children born to women with missing information on the father's death status and fertility data in both waves.We also drop children with missing heightfor-age z-scores and with different intra-household ranks across the two waves (to ensure that we were tracking the same children). 17We further restricted the sample of children to those aged under 4 years in IHDS 1 and under 10 in IHDS 2. This is because the child had to young enough so that the nutritional investments made by the mother would determine height-for-age outcomes.Our final sample consists of 44 children whose height-for-age z-score was tracked across the two waves.

Empirical implementation
Our empirical estimation consists of two parts.First, we study the effect of the early amendments in the states of Andhra Pradesh, Karnataka, Maharashtra, and Tamil Nadu on fertility outcomes of women and height-for-age z-scores of children.We exploit state-level variation in the amendments' implementation in a difference-in-differences framework using IHDS 1 to estimate these effects.Second, we examine the effect of the federal amendment legislated in 2005 on the fertility outcomes of women and height-for-age z-scores of children across India.We do this by constructing panels of women and children between 2005 and 2012 using IHDS waves 1 and 2. We define a federal amendment treatment variable based on information about women's fathers' and children's grandfathers' deaths after 2005.
14 Seventy-one percent of the households in the IHDS sample have one eligible woman.On average, households with more than one eligible woman have higher incomes relative to households with one eligible woman.The sample of households with more than one eligible woman has a higher share of landowning households, higher share of rural households, and lower share of scheduled caste/tribe households compared to the sample with just one eligible woman.However, there is no difference in the likelihood of treatment between the two samples. 15Studies have shown that this measure of early-childhood stunting explains outcomes such as cognition and earning potential later in the life of the individual (see Jayachandran and Pande 2017). 16Alternatively, matching mothers across two waves and identifying their children gives us a larger sample of children.However, in several cases, the height-for-ages and other health information for these children are recorded in only one of the survey waves.We had to then only keep the children that had recorded height-for-ages data in both the waves.Ultimately, we end up with the same sample of children. 17Any potential selection in missing records of health data could bias our results.We did not find any systematic pattern in children whose height-for-age data is missing.The survey documentation also does not provide any explanation for missing records.

Earlier state-level amendment effects
Our empirical specification for estimating the effect of the amendments passed in the four states of Andhra Pradesh, Karnataka, Maharashtra, and Tamil Nadu closely follows the difference-in-differences strategy used by Bose and Das (2020).However, our analysis is different in three ways.First, we use IHDS 1 data which has a smaller sample of women.Second, we estimate our models using district-level variation as opposed to state-level variation used by Bose and Das (2020).The finer geographic scale of analysis eliminates time-invariant factors within a state at the district level in our specification.Finally, we include several controls to account for contraception use and abortion behavior among women that are omitted in Bose and Das (2020).
The fertility analysis consists of a sample of 11,624 women surveyed in 2004-05 in IHDS 1.To estimate the effect of the state-level HSA amendments on fertility outcomes, we define a treatment dummy variable that is equal to one if a woman was unmarried at the time of the amendment in the corresponding state of legislation. 18 We include district and year-of-marriage dummies to account for temporal and timeinvariant factors that might affect the fertility outcomes of women.The empirical equation can be written as follows: where Y ikt represents fertility outcomes of woman i living in district k and married in year t; x ikt is a vector of covariates like abortion behavior, age, education, caste, contraceptive use, health, income, and landownership that also affect fertility outcomes of women; and δ k and m t denotes district and year-of-marriage dummies. 19is the error term.U nmarried ikt represents a woman i in district k and time t who was unmarried at the time of reform; Amendment State is a dummy taking value 1 if the woman resides in a state that amended the law.The coefficient α 3 is effectively a difference-in-differences estimator.It provides the magnitude of the effect of the statelevel HSA amendments on fertility outcomes by comparing unmarried and married women in amendment states and non-amendment states.
To estimate the effect of the state-level HSA amendments on height-for-age z-scores of children, we create a treatment dummy variable that takes the value one if a child's 18 We use the woman's state of residence at the time of the survey to determine whether she would be affected by the state amendments.If a positive wealth shock leads to out-of-state migration or out-of-state marriage, the results using the current state of residence for defining the treatment group may be biased.The IHDS does not record the woman's natal home state.However, out-of-state migration and out-of-state marriage has historically been very low in India.The number of women migrating across states post-marriage is a small fraction of the total number of women migrating post-marriage.According to the Census 2001 data, the number of women who migrated across states for marriage before 1991 was 8 million, whereas the number of women who migrated within states was 103 million.In other words, only about 7% of women who migrated for marriage moved across state boundaries.Hence, defining our treatment group based on the woman's state of residence at the time of the survey is unlikely to bias our results. 19Since the earlier state amendments covered all districts in a state, the state of residence is equivalent to living in one of the districts in the state.mother was unmarried at the time of the amendment in the corresponding state of legislation.The empirical equation can then be written as follows: where Z i jkt represents the height-for-age z-score of child i born to a woman married in year t in household j residing in district k; the coefficient β 3 is the differencein-differences estimator.x i jkt is a vector of covariates that include household-level controls like the share of women in the household, caste, income, and landownership, and children-level controls like gender, education, parents' education, vaccination status, and open defecation status; δ k and m t denotes district and mother's year-ofmarriage dummies. 20 There is an identification issue here.The treatment characterized by the state-ofresidence and the marital status of women living in amendment states at the time of legislation is an imperfect one.We do not observe actual inheritance of property by women.Hence, the treatment variable defined in Eqs. 1 and 2 is only a partial one, and we therefore refer to treated women as those potentially affected by or benefiting under the legislation.

Federal amendment effect
To estimate the effect of the federal amendment to the HSA legislated in 2005, we construct two panel datasets consisting of women and children, respectively.The first dataset consists of 378 married women surveyed in IHDS 1 in 2004-05 and in IHDS 2 in 2011-12.The second dataset consists of 44 children surveyed in each wave.These children, born to a subset of the 378 women comprising the first dataset, had non-missing records for height-for-age z-scores for both waves.
For both the IHDS 1 and 2, the survey records whether the woman's father had died.Using this information, we define a dummy variable Fathers Death it , which takes the value of one if a woman's father died between the two survey rounds.The suffix t represents the IHDS wave.The federal amendment covered all women, unlike the earlier state-level amendments that were applicable to only unmarried women.Hence, the treatment defined by Fathers Death it identifies the incidence of the federal amendment to the HSA.Since the federal amendment was legislated around the time of completion of the IHDS 1 survey, the federal HSA amendment would be applicable to all women whose fathers died between the two surveys.2013) also use death of the father as a treatment to test the effect of state-level amendments.Since these amendments pertain to inheritance of ancestral property, a valid question to ask is whether the father of the woman is also the eldest male member of her natal household (and thus the owner of the ancestral property).However, we do not have enough information about women's natal households to ascertain this.
The empirical equation to estimate the effect of the federal amendment on fertility outcomes can be written as follows: where Y it represents fertility outcomes of woman i in survey round t; Fathers Death it is a dummy taking value 1 if the father of woman i died between the two rounds of the IHDS; x it is a vector of time-variant covariates like abortion behavior, contraception use, health, and household income that also affect fertility outcomes of women; and θ i and τ t denote woman and time fixed effects.All time-invariant characteristics of a woman are captured by θ i , and the dummy variable τ t accounts for temporal factors that affect fertility outcomes.υ is the error term.We do not include district fixed effects because these women were living in the same location during both the survey rounds.
We use the same information on the death of the father of a woman to identify the effect of the federal amendment to the HSA on her children's height-for-age z-score outcomes.We define a dummy variable Grand f athers Death jt that takes the value of one if a child's grandfather died between the two survey rounds.The empirical equation in this case can be written as follows: where Z jt represents height-for-age z-score of child j in survey round t; x jt is a vector of time-variant covariates that include education, mother's education, father's education, share of women in the household, and household income; and ρ j and τ t denote child and time fixed effects.All time-invariant characteristics of a child are captured by ρ j .
There is an identification issue to be noted here.We do not have any information on actual property inherited by women.This implies that the treatment defined by the death of the woman's father between the two survey rounds only partially captures the effect of the legislation.

Results
The empirical results of our analysis are presented in two sub-sections.The first section presents results for the state-level HSA amendment, and the second section shows the results for the federal HSA amendment.Both sections in turn consist of two parts discussing the effect of the amendments on women's fertility outcomes and the effect on height-for-age z-score outcomes of children.

State-level amendment effect estimates
We estimate the effect of state-level amendments to the HSA on fertility outcomes of affected women and height-for-age z-score outcomes of children born to such women using the identification strategy given by Eqs. 1 and 2, respectively.Our findings are two-fold.First, consistent with Bose and Das (2020), we find that the state-level amendments led to an increase in the number of children per woman.Second, the amendments had no effect on height-for-age z-score outcomes of children born to potentially affected mothers.These results are shown in Tables 3 to 6.
Table 3 shows the effect of the amendments on the number of children born to potentially affected women.We find that, controlling for women-and householdspecific factors, women who were living in the four early-amendment states and were unmarried at the time of legislation had on average 0.38−0.41more children than others.Our results, therefore, confirm the stopping-rule mechanism discussed by Bose and Das (2020) that explains how women are able to achieve their desired fertility outcomes.
Among the different sub populations, we find that the effect of the property rights provision on fertility is strongest in rural households.The differential access to ultrasound technologies in urban and rural areas could explain why fertility rates are higher in rural areas, where women are more likely to rely on the stopping rule.We also see that the effects are slightly lower for the sample of rural households owning land compared to the overall sample of rural households.
In Appendix Table 1 we show that the effect of the legislation is stronger for rural women belonging to non-landowning households compared to rural women who belong to landowning households.This may be explained as follows.Given that the woman in the reform state is legally entitled to inherit ancestral property belonging to her natal household, even if she does not actually inherit property, she could possibly be compensated by her natal household in some way that increases her bargaining power within her marital household.On average, treated women in the sub-sample of rural non-landowning households have lesser education and lower incomes compared to rural landowning households.Hence, women in these households are less likely to have access to abortion procedures and hence more likely to attain their desired fertility through the stopping mechanism.Women living in urban areas who were affected by the legislation also had more children than other unaffected women in urban areas.This indicates that the effect of property rights provision is significant in Indian cities as well, a fact somewhat ignored by previous literature.
We also find that age has a positive and concave relationship with fertility, consistent with the discussion in Bose and Das (2020).Both the woman's and her spouse's education have a negative effect on fertility.Surprisingly, using contraceptives leads to more children per woman.This could be due to the fact that discussion around contraception is taboo in most parts of India which could potentially lead to a measurement error due to reporting bias. 22f the amendment resulted in lower child mortality, using number of children could bias our results as the number of surviving children would be higher for women in the treatment group relative to the control group.To address this, we also estimate the baseline regression using the number of children born as the outcome variable.The results, presented in Appendix Table 3, are similar to our baseline results.
Table 4 shows that the effect of the amendment is the strongest among women in households with asset ownership below the median.In fact, the effect more than

Note:
Each column presents results from a difference-in-differences regression using a sample of Hindu women surveyed in IHDS 1 during 2004-05.The dependent variable is the number of children in columns ( 1)-( 4) and the age-adjusted standardized z-score of the number of children in column ( 5).The key independent variable is the unmarried woman in amendment state dummy which takes the value one if a woman lives in an amendment state and was unmarried at the time of her resident state's amendment legislation.Columns (1) and (5) present results using the full sample of women.Columns (2), (3), and (4) present results obtained using subsamples of rural households, rural land owing households, and urban households.Covariates include a woman's age, age squared, education, and spouse's education, dummy variables for whether the woman had an abortion and uses contraceptives, the share of women in the household, log of income of the household, and dummy variables for whether the household is SC/ST, owns land, and lives in an urban area.All regressions include a woman's health status dummies, district of residence dummies, and year-of-marriage dummies.doubles in the first two quartiles relative to the last two.Anukriti et al. (2022) show that the stopping rule is more likely to be employed by households belonging to lower socioeconomic strata.This could explain why we see the highest increase in number of children among the poor.
Since the state-level analysis uses a difference-in-differences strategy, the staggered timing of treatment across states could potentially bias our results (see (Goodman-Bacon 2021)).To address this, we also estimate the baseline regression separately for each individual state.The results, presented in Appendix Table 8, show that the impact on the number of children is positive and significant.
In Table 5, we show that the state-level HSA amendments led to an increase in the number of sons and daughters born to potentially treated women.The effect on the number of sons born was higher than the effect on the number of daughters for the full sample.While the legislation led to roughly 0.20−0.29 more sons, the number of daughters born increased by roughly 0.16−0.23.This again is consistent with the stopping rule hypothesis discussed by Bose and Das (2020).However, for the urban subgroup, the effect size for the number of daughters was higher than for the number of sons.
The results in Table 6 show that the state-level amendment to the HSA had no effect on the height-for-age z-score of children born to potentially treated women.Consistent with Becker (1960) and Rosenzweig and Evenson (1977), this indicates that the additional resources received by women were expended in more children and, hence, there was no increase in resources spent on quality per child.

Federal amendment effect estimates
We estimate the effect of the federal amendment on fertility outcomes of women who potentially gained rights to a share of the family property and on height-for-age zscore outcomes of their children using the two-way fixed effects identification strategy given by Eqs. 3 and 4. Women whose fathers died between the two survey rounds are identified as being likely to benefit from the federal amendment.We find that these women had fewer children and the children born to them had better height-for-age z-score outcomes.The results are shown in Tables 7 to 10.
Table 7 shows that women whose fathers died between the two IHDS waves had 0.15−0.25 fewer children compared to those whose fathers were still alive.This is contrary to our results from Table 3.When we restrict the sample to those women who were married before the first round of the survey, and hence before the federal legislation, we see that the effect holds true (column (5) in 7).This implies that the effects of the legislation on fertility outcomes are not due to compensating mechanisms like higher education and dowry.The effect of the amendment is the highest for rural women in landowning households.The effects are robust to a specification with ageadjusted fertility levels of women.Surprisingly, contraceptive use is associated with a higher fertility among women.One possible reason could be reporting bias induced measurement error as discussed in Section 5.1.However, given that this is a panel specification, there is also a possibility of reverse causality in the relationship between contraception and fertility.This is because women who had already reached their

Note:
Each column in panels (a)-(c) presents results from a difference-in-differences regression using a sample of Hindu women surveyed in IHDS 1 during 2004-05.The dependent variable in all regressions is the number of children.The key independent variable is the unmarried woman in amendment state dummy which takes the value one if a woman lives in an amendment state and was unmarried at the time of her resident state's amendment legislation.Panel (a) presents results using the full sample of women, whereas panels (b) and (c) present results obtained using subsamples of rural households and urban households respectively.Within each panel, each column represents the corresponding subsample quartile of asset distribution in which the woman's household belongs.Covariates include a woman's age, age squared, education, and spouse's education, dummy variables for whether the woman had an abortion and uses contraceptives, the share of women in the household, log of income of the household, and dummy variables for whether the household is SC/ST, owns land, and lives in an urban area.All regressions include a woman's health status dummies, district of residence dummies, and year-of-marriage dummies.Panel (a), panel (b), and panel (c) regressions with full controls are available in Appendix Tables 5, 6, and 7

Note:
Each column in panels (a) and (b) presents results from a difference-in-differences regression using a sample of Hindu women.One woman per household was surveyed in IHDS 1 during 2004-05.In panel (a), the dependent variable is the number of sons in columns ( 1)-( 4) and the age-adjusted standardized z-score of the number of sons in column (5).In panel (b), the dependent variable is the number of daughters in columns ( 1)-( 4) and the age-adjusted standardized z-score of the number of daughters in column ( 5).The key independent variable is the unmarried woman in amendment state dummy which takes the value one if a woman lives in an amendment state and was unmarried at the time of her resident state's amendment legislation.In both panels (a) and (b), columns ( 1) and ( 5) present results using the full sample of women, whereas columns (2), (3), and (4) present results obtained using subsamples of rural households, rural land owing households, and urban households.Covariates include a woman's age, age squared, education, and spouse's education, dummy variables for whether the woman had an abortion and uses contraceptives, the share of women in the household, log of income of the household, and dummy variables for whether the household is SC/ST, owns land, and lives in an urban area.All regressions include a woman's health status dummies, district of residence dummies, and year-of-marriage dummies.Panel (a) and panel (b) regressions with full controls are available in Appendix Tables 9 and  10 The dependent variable is the number of children in columns ( 1)-( 5) and the age-adjusted standardized z-score of the number of children in column ( 6).The key independent variable is the father's death dummy which takes the value one if a woman's father died between the two survey waves and zero otherwise.Columns ( 1) and ( 6) present results using the full sample of women.Columns ( 2), ( 3), (4), and (5) present results obtained using subsamples of rural households, rural land owning households, urban households, and women who were married before the first wave of IHDS.Covariates include a woman's education and spouse's education, dummy variables for whether the woman had an abortion and uses contraceptives, the share of women in the household, and log of income of the household.All regressions include a woman's health status dummies, woman fixed effects, and survey wave fixed effects.Regressions without controlling for contraceptive use are available in Appendix desired number of children would be more likely to use contraceptives than those who had not.In Appendix Table 11, we show that the results without the contraceptive dummy are similar but of a higher magnitude compared to those seen in Table 7.
Since the federal amendment covered women living in all states, women living in states with an earlier amendment to the HSA might have different outcomes than those living in states that were only treated by the federal amendment.The effect of the federal amendment could strengthen the institutional reforms already in place in the early-amendment states for decades.On the other hand, the gains in such states could be muted by the fact that women living there expected inheritance or other forms of compensation and the federal reform did not change that expectation.As noted earlier, women who were married before the state amendments, and hence excluded from inheritance, became eligible to inherit ancestral property under the federal amendment.In Table 8, we show that the federal amendment impacted fertility outcomes for women living in the states that only had a federal amendment.The impact of the federal amendment is not statistically significant for women living in the states that had already amended the HSA.These results indicate that the effect of the amendment on fertility is stronger among women living in late-amendment states, whereas fertility outcomes for women in early-amendment states were not affected by the federal amendment.
Table 9 confirms that women with more resources due to better property rights provision had fewer sons.Hence, the decline in the number of children as a result of the federal amendment can be entirely explained by the decline in the number of sons.While there was no effect of the legislation on the number of daughters, the number of sons born to treated women declined by 0.14−0.19.The effect is significant for women in rural areas and women married before the legislation.
In Table 10, we show that the federal amendment increased the height-for-age zscore of children by 1.8−3.3standard deviation.For the full sample (in column 1 Table 10), the coefficient is 2.28 and is significant at 1%.This result is consistent with a quantity-quality trade-off.Children living in urban areas saw the maximum amount of gain in height as a result of the federal reform.Better healthcare in urban areas could explain the gain from the reform.

Mechanisms
In this section, we discuss the specific channels following the inheritance reforms that could explain our results.The first channel is a positive wealth shock through the inheritance of property.Land and property are considered to be the most valuable assets for households and can be used either as an additional income streams or as collateral to borrow money or to generate revenue through outright sale.This could translate into additional resources for women and empower them to realize their fertility outcomes as well as invest in their children.However, as mentioned in Section 4, our empirical strategy cannot directly identify whether women inherited property after the amendment.An indirect proxy for determining whether the woman is likely to inherit property is to examine whether the fertility impact is heterogeneous with respect to the number of siblings.If having a sibling reduces the likelihood of  Note: Each column in panels (a) and (b) presents results from a two-way fixed effects regression using a panel of Hindu women surveyed in IHDS 1 during 2004-05 and IHDS II during 2011-12.Panel (a) restricts the sample to those women living in states that did not have an earlier state-level amendment to the HSA.Panel (b) restricts the sample to those women living in states that had an earlier state-level amendment to the HSA.In both panels (a) and (b), the dependent variable is the number of children in columns (1)-( 5) and the age-adjusted standardized z-score of the number of children in column ( 6).The key independent variable is the father's death dummy which takes the value one if a woman's father died between the two survey waves and zero otherwise.In both panels (a) and (b), columns ( 1) and ( 6) present results using the full sample of women, whereas columns (2), ( 3), (4), and (5) present results obtained using subsamples of rural households, rural land owing households, urban households, and women who were married before the first wave of IHDS.Covariates include a woman's education and spouse's education, dummy variables for whether the woman had an abortion and uses contraceptives, the share of women in the household, and log of income of the household.All regressions include a woman's health status dummies, woman fixed effects, and survey wave fixed effects.Panel (a) and panel (b) regressions with full controls are available in Appendix Tables 12 and 13   In panel (a), the dependent variable is the number of sons in columns ( 1)-( 5) and the age-adjusted standardized z-score of the number of sons in column (6).In panel (b), the dependent variable is the number of daughters in columns ( 1)-( 5) and the age-adjusted standardized z-score of the number of daughters in column (5).The key independent variable is the father's death dummy which takes the value one if a woman's father died between the two survey waves and zero otherwise.
In both panels (a) and (b), columns ( 1) and ( 6) present results using the full sample of women, whereas columns (2), ( 3), (4), and (5) present results obtained using subsamples of rural households, rural land owing households, urban households, and women who were married before the first wave of IHDS.Covariates include a woman's education and spouse's education, dummy variables for whether the woman had an abortion and uses contraceptives, the share of women in the household, and log of income of the household.All regressions include a woman's health status dummies, woman fixed effects, and survey wave fixed effects.Panel (a) and panel (b) regressions with full controls are available in Appendix Tables 14 and 15 inheriting property or leads to a much smaller share in the property, we would see an attenuated effect of the treatment on the outcome variables.Using data on the number of siblings, we interact the treatment variable with the number of siblings and estimate the effect on the number of children.The results for the state amendments and federal amendment are presented in Appendix Tables 16 and 17, respectively.For the state amendments, the coefficient for number of children is smaller for potentially treated women with more siblings than for potentially treated women with no siblings.For the federal amendment, the interaction term between treatment and number of siblings is not significant.In other words, the number of siblings that a potentially treated woman has does not affect the number of children.
According to the Agricultural Census, the share of total landholdings and area of landholdings owned by women in India went up from 9.5 to 11.7% between 1995-96 and 2005-06.Thus, while the share of women's ownership of agricultural land holdings increased over time, the share in total ownership is still extremely low.Moreover, if such land is non-productive or if women have no de facto control over its use despite being the owner, then a wealth effect would not be realised.Furthermore, from prior literature on the state-level amendments to the HSA, there is some ambiguity regarding actual inheritance of land.Deininger et al. (2013) find that the likelihood of inheritance increases among women after the state-level amendments.On the other hand, studies indicate that household heads were more likely to give away joint family land as gifts to their sons to evade the legislation (Roy 2015;Sivaramayya 1988).Agarwal et al. (2021) finds that widows are more likely to inherit land relative to daughters.Thus, this mechanism of a positive wealth effect can only partially explain the fertility outcomes.
The second channel is through increased intra-household bargaining power of women impacted by the legislation.This can happen in two ways.First, as discussed by Roy (2015), households evading the inheritance rights legislation compensated daughters with more education and dowry.Given that agriculture in India has been impacted by declining returns while returns to education are high, if such compensation leads to employment in non-farm sectors, it would have similar effects on fertility of women and height-for-age z-score of their children as a direct wealth shock.A number of studies have documented positive returns to education in the form of higher earnings.Thomas (1994) finds that in the USA, Brazil, and Ghana, maternal education has a larger effect on heights of girl children relative to male children.Estudillo et al. (2001) find that while sons were more likely to inherit land in rural Philippines, parents tended to invest more in the schooling of female children.They show that education and land transfers were close substitutes as means for transferring wealth to the next generation.
Second, the inheritance law could create some expectation that a woman will be likely to acquire property through her natal household in the future, which would increase her bargaining power within her marital household.This would allow her to claim more resources and realize her desired fertility outcomes.We can potentially test this in the case of early state amendments. 23We examine whether the reforms increased women's bargaining power with two outcome variables: a dummy variable taking value 1 if the woman's name is on the family bank account and a dummy variable taking value 1 if the woman's name is on the ownership documents of the home.We further examine whether potentially treated women were able to realize their desired fertility outcomes.For this, we create a dummy variable that takes value 1 if the woman reported desiring more children and 0 otherwise.The results, presented in Appendix Table 18, show that women who were potentially affected by the state amendments were more likely to be joint owners of their homes and were less likely to desire more children.

Additional fertility indicators
Our main empirical analysis compares the number of children born to potentially treated and untreated women in the aggregate.The IHDS also records retrospective birth histories of the women in the sample.This allows us to create a panel of women starting from the year of marriage and use an outcome variable that takes value 1 in the year that a woman had a child and 0 in all other years.This probability of birth is an alternate indicator of fertility that exploits the granular variation in the timing of birth before and after the reforms.The empirical specification is a two-way fixed effects regression with women and year fixed effects.The main independent variables for the state and federal reforms are the same as described in Eqs. 1 and 3 respectively.
Table 11 presents the results for this new measure of fertility.Consistent with the results in Table 3, we see that the state amendment is associated with an increase in the probability of birth for women who were unmarried in the reform state at the time of the reform.The reform was associated with an increase in the probability of birth of both sons and daughters, with the effect being higher for daughters than sons.This is contrary to the results in Table 5.One possible explanation for this could be that with potentially treated women exercising a stopping rule to achieve son preference, we would see probability of daughters being born go up.With the federal reform, potentially treated women had lower probability of birth and a lower probability of birth of both sons and daughters.Consistent with Table 9, the effect on the probability of sons was negative and significant.
In summary, the fertility results using probability of birth and probability of sons are consistent with results using number of children and number of sons for both state and federal amendments.However, in the case of state amendments, the effect is stronger for probability of daughters compared to sons (but stronger for number of sons compared to number of daughters), and in the case of the federal amendment, the effect on probability of daughters is negative and significant.

Additional quality indicators
Any changes in nutritional resources to children as a result of the amendment would also be captured in weight-for-age z-scores.Hence, we estimate Eqs. 2 and 4 using weight-for-age z-scores of children as an alternate outcome variable to measure quality.The results are presented in Table 12.The state-level amendment had no effect on the weight-for-age z-score of children born to women potentially affected by the legislation.On the other hand, children born to women potentially affected by the federal amendment had a 0.67 higher standard deviation of weight-for-age z-score on average.Children living in rural landed households had the highest increase.Thus, our results are consistent with the full sample results in Tables 6 and 10, when using weight-for-age z-score as a measure of quality.
An increase in access to resources for women due to the amendment could also potentially affect child mortality.To test this, we use a probit regression where the outcome variable is a dummy that takes value 1 if the woman in the sample gave birth to a child that later and 0 otherwise.The independent variables used in the state-level and federal amendment regression specifications are the same as in Eqs. 1 and 3. We find that neither the state nor federal reforms had an impact on this outcome.

Why do state and federal amendment effects differ?
The estimates for the effect of the state and federal reforms on fertility outcomes presented in Tables 3 and 7 respectively are of opposite signs.In this section, we examine potential methodological, theoretical, and institutional factors that could account for this divergence.
A major reason for the divergence is the fact that the identification strategies are different in the two instances.While treatment status due to state amendments is based on state-of-residence and the marital status of women at the time of legislation, the treatment capturing the federal legislation is a more direct one based on the death of the woman's father. 24Besides, the sample sizes in the analyses to estimate the effect of the state and the federal legislation are different.The low sample sizes in the case of the federal amendment regressions explain the lack of significance in some of the estimates in Tables 7 to 10.
Even if we were to account for the identification issues discussed above, we might see a difference in the state and federal amendment effects on fertility and height-forage z-score outcomes.The theoretical framework implies that the empirical effects could go in either direction and could also differ based on households' socioeconomic status.Rosenzweig and Evenson (1977) show that increase in landholdings could push up fertility rates among rural agrarian households as more children are needed to work on farms.In a similar vein, we may see a stronger positive relationship between increase in wealth and fertility among poorer households.This is confirmed when we compare the coefficients for the quantity of children and treatment effect of state amendments for households across different asset quartile groups.We see that the positive relationship is strongest for households with assets below the median in rural areas.Another potential reason for an increase in the number of children is due to women employing a stopping rule where they have more children to attain the desired number of sons.Again, there are disparities in the use of the stopping rule across income classes, with the poor more likely to use it whereas the rich rely on sexselective abortions (Anukriti et al. 2022).
On the other hand, drawing from Becker and Lewis (1973), an increase in wealth could lead to a fall in quantity but a rise in the quality of children.This qualityquantity trade-off may be more prevalent among the non-poor who have the means to be more effective in reaching the desired fertility and in using the additional resources to improve nutrition for children.Furthermore, when cost of childcare is disproportionately borne by women and when the motivation to have children is for old age security, an increase in women's autonomy could result in women having fewer children (Eswaran 2002).
Second, access to sex determination through ultrasound and abortion technology as well as the laws governing sex determination may have been very different during the respective time periods in question.Abortion has been legalized in India since the 1970 s, and pre-natal ultrasound technology has been available since the 1980 s (Anukriti et al. 2022).However, while access to abortion processes was available under both periods, factors such as drastic lowering of import tariffs and increased domestic production of ultrasound scanners improved access to sex determination around the time of the federal amendment in 2005 relative to the time of the state amendments. 25f the norm of son preference remained unchanged over time, the opposite signs with respect to the number of children under the state and federal amendments could be a result of women using increased bargaining power and resources due to the state amendment to use the stopping rule (having more children until the desired number of sons was reached) and women using increased resources due to the federal amendment to access sex determination and sex-selective abortion to have the desired number of sons (and hence having fewer children).However, this does not explain the opposite signs with respect to the number of sons under the state and federal amendments.The fact that women potentially affected by the federal amendment had fewer sons possibly suggests that cultural norms around son preference have weakened and that the inheritance laws influenced women's ability to take care of themselves in the future without relying on sons. 26hird, even though the state and federal amendments were similar in that both were meant to provide women with property rights, there are differences.The first point of difference is the geographical overage of the laws; state amendments were only applicable within the South Indian states of Andhra Pradesh, Maharashtra, Karnataka, Tamil Nadu, and Kerala, whereas the federal amendment was applicable to all states.The south Indian states have different socioeconomic conditions compared to the rest of the Indian states.The north-south divide in India has been discussed in the context of different strands of literature (Paul and Sridhar 2015;Sridhar and Reddy 2011).The second difference is that while the state amendments applied only to those women who were unmarried at the time of the reform, the federal amendment applied to all women.Thus, the treatment groups differed across the two reform regimes.

Conclusion
Reforms aimed at granting property rights to women have the potential to profoundly impact societies.This paper studies the impact of state and federal amendments to the Hindu Succession Act 1956.It tests whether granting land and property inheritance rights to women in India leads to a quantity-quality trade-off.
Contrasting theories suggest that the effect of the reform on fertility outcomes could go either way.Our results show that women potentially benefiting from the state amendments had more children.This is consistent with the findings of Bose and Das (2020) and could possibly be due to women realizing their desired son preference through the stopping mechanism.The state reforms had no effect on the height-forage z-scores of the children born to potentially affected women.On the other hand, women benefiting under the federal amendment had fewer children, and the children born to women in the treatment group had higher height-for-age z-scores relative to the control group.Finally, while the impact of the state amendments on the number of sons was slightly higher than on the number of daughters, women benefiting under the federal amendment had fewer sons, but the number of daughters did not change.
However, there are important caveats to these findings.We recognize that due to data limitations, we are unable to directly observe and measure land inheritance, and hence, we only identify women eligible to inherit land after the reform.Furthermore, the state-level analysis uses variation based on the state of residence, and we assume that a woman's marital and natal state of residence is the same.For the federal-level analyses, our results could be affected by low sample sizes.
The contradicting effects of the state and federal amendments can be attributed to differences in institutional factors, identification strategies, and evolving cultural norms.The mechanisms driving these results include a potential direct positive wealth shock due to inheritance and increased bargaining power of women within their marital households. 21

20
Prior literature indicates that hygiene and sanitation factors like open defecation are critical in the development of children in early years (Spears 2020). 21Deininger et al. ( column in panels (a) and (b) presents results from a two-way fixed effects regression using a panel of Hindu women surveyed in IHDS 1 during 2004-05 and IHDS II during 2011-12.

Table 1
Summary statistics for state-level amendment analysis Source: Authors' calculations based on IHDS wave 1 2004-05

Table 2
Summary statistics of panel data for federal analysis Source: Authors' calculations based on IHDS data wave 1 and wave 2.

Table 3
State amendments' effects on the number of children Regression in column (2) is subdivided by landownership and available in Appendix Table 1.All regressions are run without the contraceptive control and available in Appendix Table 2. Regressions without the spouse's education control and controlling for highest education level of male and female household members are available in Appendix Table 4. Standard errors clustered at the district level.* p

Table 4
State amendments' effects on fertility by household assets , respectively.Standard errors clustered at the district level.*p Each column presents results from a difference-in-differences regression using a sample of children born to Hindu women surveyed in IHDS 1 during 2004-05.The dependent variable is the height-for-age z-score of a child.The key independent variable is the unmarried mother in amendment state dummy which takes the value one if the mother of the child lives in an amendment state and was unmarried at the time of her resident state's amendment legislation.Column (1) presents results using the full sample of children.Columns (2), (3), and (4) present results obtained using subsamples of children living in rural households, rural land owing households, and urban households.Covariates include a child's education, their mothers' and fathers' education, share of women in the household, log of income of the household, and dummy variables for the child's vaccination status, open defecation status, whether the child is female, and whether the household of the child is SC/ST, owns land, and lives in an urban area.All regressions include children's birth order dummies, the mother's health status dummies, district of residence dummies, and mother's year-of-marriage dummies.Standard errors are clustered at the district level.* p Each column presents results from a two-way fixed effects regression using a panel of Hindu women surveyed in IHDS 1 during 2004-05 and IHDS II during 2011-12.

Table 8
Federal amendment's effects on fertility by early-amendment and late-amendment states

Table 10
, respectively.Standard errors clustered at the district level.* p Federal amendment's effects on children's height-for-age z-score Each column presents results from a two-way fixed effects regression using a panel of children born to Hindu women surveyed in IHDS 1 during 2004-05 and IHDS II during 2011-12.The dependent variable is the height-for-age z-score of a child.The key independent variable is the grandfather's death dummy which takes the value one if a child's grandfather died between the two survey waves and zero otherwise.
The independent variable is the unmarried mother in amendment state dummy which takes the value one if the mother of the child lives in an amendment state and was unmarried at the time of her resident state's amendment legislation.The dependent variable in column (1) is a dummy variable taking value 1 in the year in which the woman gave birth and 0 otherwise.The dependent variable in column (2) is a dummy variable taking value 1 in the year in which the woman gave birth to a son and 0 otherwise.The dependent variable in column(3) is a dummy variable taking value 1 in the year in which the woman gave birth to a daughter and 0 otherwise.All regressions include woman and year fixed effects.In panel (b), each observation is a woman surveyed in IHDS 1 during 2004-05 and during IHDS II during 2011-12 in year i, where year of marriage >= i <= 2005.The key independent variable is the father's death dummy which takes the value one if a woman's father died between the two survey waves and zero otherwise.The dependent variable in column (1) is a dummy variable taking value 1 in the year in which the woman gave birth and 0 otherwise.The dependent variable in column (2) is a dummy variable taking value 1 in the year in which the woman gave birth to a son and 0 otherwise.The dependent variable in column (3) is a dummy variable taking value 1 in the year in which the woman gave birth to a daughter and 0 otherwise.All regressions control for whether the woman was residing in a state that had an earlier HSA amendment and include woman and year fixed effects.Standard errors are clustered at the district level.*pPanel(a)presentsresults from a difference-in-differences regression using a sample of children born to Hindu women surveyed in IHDS 1 during 2004-05.The dependent variable is the weight-for-age z-score of a child.The key independent variable is the unmarried mother in amendment state dummy which takes the value one if the mother of the child lives in an amendment state and was unmarried at the time of her resident state's amendment legislation.Column (1) presents results using the full sample of children.Columns (2), (3), and (4) present results obtained using subsamples of children living in rural households, rural land owing households, and urban households.Covariates include a child's education, their mothers' and fathers' education, share of women in the household, log of income of the household, and dummy variables for the child's vaccination status, open defecation status, whether the child is female, and whether the household of the child is SC/ST, owns land, and lives in an urban area.All regressions include children's birth order dummies, the mother's health status dummies, district of residence dummies, and mother's year-of-marriage dummies.Standard errors are clustered at the district level.Panel (b) presents results from a two-way fixed effects regression using a panel of children born to Hindu women surveyed in IHDS 1 during 2004-05 and IHDS II during 2011-12.The dependent variable is the weight-for-age z-score of a child.The key independent variable is the grandfather's death dummy which takes the value one if a child's grandfather died between the two survey waves and zero otherwise.Covariates include a child's education, the mother's and the father's education, share of women in the household, and the log of income of the household.All regressions include mother's health status dummies, children fixed effects, and survey wave fixed effects.Panel (a) and panel (b) regressions with full controls are available in Appendix Tables19 and 20, respectively.Standard errors clustered at the